Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Mapping VMEs
First Meeting of the Protected Areas and Ecosystems Working Group (PAEWG1)
FAO workshop on vulnerable marine ecosystems (VMEs) for the SIOFA region
18-19 March, 2019
National Research Institute of Fisheries Science, Yokohama, Japan
Ashley A. Rowden
FAO DSF Guidelines and actions taken by RFMOs in other regions
Outline
• FAO Guidelines
• NAFO
• SPRFMO
• Mapping VME issues
• Recent developments and future directions
Background
• Bottom trawling impacts seafloor habitats, communities and species
• UNGA passed resolutions to protect VMEs in 2006 and 2009
• RFMOs obliged to prevent SAI to VMEs
• FAO produces guidelines in 2009 to assist RFMOs
untrawled
trawled
FAO Guidelines
The role of the Guidelines is to provide tools, including
guidance on their application, to facilitate and encourage the
efforts of States and RFMO/As towards sustainable use of
marine living resources exploited by deep-sea fisheries, the
prevention of significant adverse impacts on deep-sea VMEs
and the protection of marine biodiversity that these ecosystems
contain.
FAO Guidelines
12. In order to achieve these objectives, States and
RFMO/As should:
…
ii. identify areas where VMEs are known or likely to occur;
…
47. Flag States and RFMO/As should conduct assessments to
establish if deep-sea fishing activities are likely to produce
significant adverse impacts in a given area. Such an impact
assessment should address, inter alia:
…
iii. identification, description and mapping of VMEs known or
likely to occur in the fishing
area;
…
FAO Guidelines
FAO Guidelines
32. Sufficiently fine-scaled data are required as a basis for the
assessment of stock status and impacts on VMEs. In addition,
fishery-independent research surveys are encouraged, in particular
to provide relevant information on VMEs and how they are
affected by anthropogenic activities.
44. As a necessary step towards the identification of VMEs,
States and RFMO/As, and as appropriate FAO, should assemble
and analyse relevant information on areas under the competence
of such RFMO/As or where vessels under the jurisdiction of
such States are engaged in DSFs or where new or expanded
DSFs are contemplated.
45. Where site-specific information is lacking, other information
that is relevant to inferring the likely presence of vulnerable
populations, communities and habitats should be used.
FAO Guidelines
To avoid SAI on VMEs, RFMOS should:
• Identify areas where VMEs are known or likely to occur
➢Use data from stock assessment surveys, independent surveys, fisheries bycatch
➢ Infer distribution of VMEs where data lacking
How do you identify areas where VMEs are known or likely to occur?
Northwest Atlantic Fisheries Organisation South Pacific Regional Fisheries Management Organisation
Examples from…
NAFO
• Biomass and species richness distribution of VME indicator taxa
• Kernel density approach to identify concentrations of VME indicator taxa
• Canadian and Spanish/EU bycatch data for – sponges, corals, seapens
Murillo et al. (2011)
Sample locations
Murillo et al. (2011)
Biomass Species richness
Biomass and species richness distribution
Kernel density
Kenchington et al. (2014)
• Identifies ‘hotspots’ based on a neighbourhood approach using a spatially-defined threshold
• Software used in GIS to automate production of the polygon surfaces for range of thresholds to identify most appropriate threshold of ‘natural’ concentrations
Kernel density
Kenchington et al. (2014)
• Kernel density map showing 75 kg threshold used to define VMEs and range of other biomass thresholds
• Cross-checked against other criteria and data
SPRFMO
• Habitat suitability modellingof VME indicator taxa
• Trialled different scales of HSM
• Mostly NZ and Australian data for 10 VME indicator taxa, including corals, sponges etc Anderson et al (2016a)
Corals
Field records for species or community and maps of environment
Map of probability that species or community is present
Statistical model(including internal validation)
Habitat suitability modelling(also known as Species Distribution Modelling)
Habitat suitability modelling
• SPRFMO-scale models
• Ground truth validation revealed poor performance
• Mostly related to inaccurate global bathymetry
Anderson et al (2016a)
Habitat suitability modelling
• NZ-regional models
• Performed better (internal validation only) using regional bathymetry
• Mapped model uncertainty
• But map did not include all areas of interest
uncertainty
Goniocorelladumosa
Anderson et al (2016b)
Habitat suitability modelling
• SW Pacific-scale models
• Three types of models -RF, BRT, MaxEnt – for each VME indicator taxon
• Also mapped model uncertainty
Georgian et al (2019)
Habitat suitability modelling
• Many different types of HSM models – which should you use?
• Ensemble approach combines models by averaging output weighted by individual model performance
MaxEnt
RF
BRT
Ensemble
Solenosmilia variabilis
Georgian et al (2019)
Habitat suitability modelling
Georgian et al (2019)
• Calculated CV (from bootstrapping) as metric of model uncertainty –projected spatially
• There are other ways to measure uncertainty
• Important to do
Solenosmilia variabilis
Ensemble
EnsembleSpatial CV
Finally….
Acknowledge the fundamental issue with models but emphasise their purpose
“Essentially, all models are wrong, but some are useful.”
George Box
Mapping VME issues
• Are the mapped VMEs connected or isolated? –need to know the likely extent of connectivity among VMEs in different areas
e.g., VMEs in some areas will be more or less important for providing recruits to sustain overall population in region, and recovery in disturbed areas – and thus more important to protect from trawling impacts
?
Recent developments and future directions
Determining VME connectivity
• NAFO area study
• Determine connectivity of VME indicator taxa among closed areas to assess their effectiveness
Kenchington et al. (2019)
Recent developments and future directions
Determining VME connectivity
• Biologically parametrised particle tracking models
• Determine spatial and temporal dispersal paths of theoretical larvae from closed areas
Kenchington et al. (2019)
Surface water 100 m
Recent developments and future directions
Determining VME connectivity
• Hindcast dispersal models to assess the source of larvae for the the closed areas
• Potential source populations inferred from habitat suitability models
Sponges
Kenchington et al. (2019)
Recent developments and future directions
Determining VME connectivity
• Dispersal models indicate a degree of connectivity among closed areas, with some areas being key suppliers of recruits
• Hindcasting indicates that some recruitment likely from VMEs outside of closed areas
Kenchington et al. (2019)
Mapping VME issues
• Are we actually mapping VMEs? – need to identify abundance or biomass thresholds that relate to the FAO’s VME functional criteria
e.g., structurally complex VMEs should be “created by significant concentrations of biotic and abiotic features”, and that “such ecosystems often have high diversity, which is dependent on the structuring organisms”
Recent developments and future directions
Identifying thresholds
• Seamount-scale models
• 25 m resolution
• Based on image records and bathymetric variables derived from MBES data
Rowden et al. (2017)
Recent developments and future directions
Identifying thresholds
• Ensemble P/A and abundance models
• Applied a subjective/expert density threshold to abundance models to identify coral reef VMEs
Solenosmilia variabilis
Rowden et al. (2017)
Recent developments and future directions
Identifying thresholds
• Application of threshold identifies only small areas of patch reefs
• But these VME maps depend on the veracity of threshold definition
Coral reef VME
Rowden et al. (2017)
Recent developments and future directions
Identifying thresholds
• A later comparison of P/A and abundance models indicates that perhaps threshold is ok
• New work trying to establish link between abundance and function of VME (i.e. elevated biodiversity)
Rowden et al. (in prep.)
Spe
cies
ric
hn
ess
Number of coral heads
Pro
bab
ility
of
HS
Modelled data for Forde
Recent developments and future directions
% Mud
• Improve mismatch between scale of environmental predictors and biological records/response
• Incorporate uncertainty in environmental predictor variables in models
Recent developments and future directions
• Model and map recovery potential of VMEs
• Predict and map effect of future climate change on HS for VMEs
Anderson et al. (2015)
Thank you